IU MolViz User Guide

Insight II Notes: Docking


Manual Docking with a Grid

Ideally, we would like to move the interacting molecules in real time on the screen of a workstation while computing the interaction energy. While the energy expression is straghtforward to compute, the computation time increases as the square of the number of interacting atoms, making the process too slow for many molecular systems. An energy grid approximating the larger of the two molecules can be precomputed. Since the interaction energy can then be approximated by calculating the energy between atoms of the moving molecule and the appropriate grid points, the docking can be accomplished much faster.

Docking with a grid was very popular during the 1980s when computers were not very powerful. Even though computing power has improved during the last decade, the grid method is still fairly popular, unless both molecules are relatively small. Use the Docking Module in InsightII to create a grid and dock molecules.


Traction Beam Docking

If you know that two specific atoms of each molecule must interact (i.e., there is a hydrogen bond, a salt bridge, or an important van der Waals contact that must be present in the final complex), you can apply the "traction beam method" of docking. This method is just your basic energy minimization or molecular dynamics with one molecule held fixed and a low-force-constant distance restraint applied between the two atoms. When the non-fixed molecule is minimized, it's final conformation must also satisfy the distance restraint, so this molecule slowly "swims" towards the fixed molecule until the distance restraint is satisfied.

To set up the docking methods described below:

  1. Build your structure using the Builder or Biopolymer modules in InsightII. Or get a PDB-style molecular coordinate file.
  2. Using Forcefield/Potentials in Builder or Biopolymer, FIX your potentials, FIX your partial charges, and FIX your formal charges. If you get no error messages, go to the Discover module. If you get error messages, you must manually fix your molecule. See the Building molecules in InsightII page for details and tips.
  3. Using the Constraints menus, fix regions of your molecules that should not change conformation, add distance restraints between pairs of atoms, add torsion restraints, tether atoms in regions of your molecule that should shange ther coordiantes by a small amount, etc.
  4. Click on Run/Run, and select the local machine, the molecule name, COMMAND FILE, auto-assign parameters, minimize, and Reduce Output. Then click EXECUTE

To start the docking methods described below:

  1. Set up the docking in a UNIX window:
  2. Type ls. You should have three files, named moleculename.car (the input coordiantes), moleculename.mdf (the molecular data file, containing info about bonding, charges, etc.), and moleculename.inp (the "input" instructions for the simulated annealing calculation).
  3. Edit the moleculename.inp file so that it contains appropriate "input" instructions for simulated annealing. See the example input files in the methods described below..
  4. Type discover
  5. Delete the moleculename.his file. You don't need it, and it eats up mondo disk space.
  6. To determine if your calculation is proceeding,
  7. Again, make sure that you have deleted the moleculename.his file. You don't need it, and it eats up mondo disk space.
  8. Log out
  9. After a while, log back in.
  10. To determine if the simulated annealing is finished:

Minimization Strategies

These minimization strategies work well only if the force constant of the distance restraint is very low. This allows the ligand to retain it's optimized geomerty while it is dragged to the protein's surface. Since these methods don't allow the ligand to change it's conformation by a significant amount, these methods are a "lock and key" docking method, where the lock and key don't change shape.

Using a constant-force traction beam

See the min_cftb.inp input file.
This method minimizes the ligand using the standard forcefield, with an extra distance restraint between one ligand and one protein atom. The final complex must satisfy this distance constraint, so that the ligand will be "bound" to the protein.

Using a force-ramped traction beam

See the min_frtb.inp input file.
This method minimizes the ligand using the standard forcefield, with an extra distance restraint between one ligand and one protein atom. The force constant of the distance restraint is increased during the minimization. This allows the ligand to be relatively unrestrained by the distance restraint during the first parts of the minimization; during the end of the minimization, the distance restraint must be satisfied. Since the final complex must satisfy this final distance constraint, the ligand will be "bound" to the protein.

Using a variable-length traction beam

See the min_vltb.inp input file.
This method minimizes the ligand using the standard forcefield, with an extra distance restraint between one ligand and one protein atom. The value of the restraining distance is decreased during the minimization. This allows the ligand to be restrained within a large distance from the protein during first parts of the minimization; during the end of the minimization, the ligand must be restrained to be very near the protein. Since the final complex must satisfy this final distance constraint, the ligand will be "bound" to the protein.


Dynamics Strategies

These strategies allow the ligand to sample a greater range of conformation space during the docking, compared to the minimization methods described above. Parts of the proten may also be allowed to move. Since the conformation of the ligand iand parts of the protein are allowed to change, these methods are a "hand and glove" docking method, where the hand and glove are allowed to change shape. These methods require more computation time.

Using a constant-force traction beam

See the dyn_cftb.inp input file.
This method performs simulated annealing with the ligand (the protein is fixed) using the standard forcefield, with an extra distance restraint between one ligand and one protein atom. The final complex must satisfy this distance constraint, so that the ligand will be "bound" to the protein. This method is technically a simulated annealing method, where-by the ligand is slowly cooled and minimized, instead of a dynamics method, where-by the ligand is NOT cooled or minimized. Simulated annealing is required in this method to force the ligand to eventually satisfy the distance constraint.

Using a force-ramped traction beam

See the dyn_frtb.inp input file.
This method performs molecular dynamics with the ligand (the protein is fixed) using the standard forcefield, with an extra distance restraint between one ligand and one protein atom. The force constant of the distance restraint is increased during the dynamics. This allows the ligand to be relatively unrestrained by the distance restraint during the first parts of the dynamics; during the end of the dynamics, the distance restraint must be satisfied. Since the final complex must satisfy this final distance constraint, the ligand will be "bound" to the protein.

Using a variable-length traction beam

See the dyn_vltb.inp input file.
This method performs molecular dynamics with the ligand (the protein is fixed) using the standard forcefield, with an extra distance restraint between one ligand and one protein atom. The value of the restraining distance is decreased during the dynamics. This allows the ligand to be restrained within a large distance from the protein during first parts of the dynamics; during the end of the dynamics, the ligand must be restrained to be very near the protein. Since the final complex must satisfy this final distance constraint, the ligand will be "bound" to the protein.


AutoDock

A Monte Carlo simulation can be performed where a small ligand is allowed to "randomly walk" about the surface of a protein or nucleic acid. This method is very computationaly intensive. Since there is some randomness to the generation of the final molecular complex, and since there are usually many protein-ligand conformations with similar energy, an ensemble of ~10 structures is usually calcualted with the hope that most of the ~10 structures will show the same binding mode. This means that this computationally intensive program must be repeated ~10 times. To perform the Monte Carlo simulation:
  1. The temperature of the system is set to a high value (~500K)
    1. The coordinates of the ligand are changed randomly
      1. the energy of interaction between the two molecules is calculated
      2. if the energy of interaction is lower than the previous step, this new conformation is ACCEPTED
      3. if the energy of interaction is higher than the previous step, the difference in energy is calculated; the probability (between 0 and 1) that the energy could increase by this amount is then calculated based upon the Boltzman distribution at the system's temperature. A random number (between 0 and 1) is chosen. If the random number is larger than the probability, then this new conformation is REJECTED; if the random number is smaller than the probability, then this new conformation is ACCEPTED.
    2. Step 1 is repeated until the number of ACCEPTED or REJECTED steps reaches a pre-set limit (e.g., 100,000 steps)
  2. The temperature is reduced by a few percent, and step 1 is repeated at this new temperature.
  3. Once the final temperature reaches a very low value (~1K), the Monte Carlo simulation is finished.

Contact the MolViz Facility Staff if you want to use AutoDock.


Back to  |  Insight II Notes  |  Application Guide  |  MolViz Home  |
Send comments to chemvis@indiana.edu
Last updated: 01/23/2001